{"id":2343,"date":"2026-09-21T10:09:16","date_gmt":"2026-09-21T01:09:16","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/2026\/09\/21\/uncensored-qwen-image-2-1-gguf-released\/"},"modified":"2026-09-22T10:20:53","modified_gmt":"2026-09-22T01:20:53","slug":"uncensored-qwen-image-2-1-gguf-released","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/21\/uncensored-qwen-image-2-1-gguf-released\/","title":{"rendered":"Uncensored Qwen-Image-2.1 GGUF Released for Local ComfyUI"},"content":{"rendered":"<p><em>Sample outputs are available on the <a href=\"https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF\">model card<\/a>.<\/em><\/p>\n<blockquote>\n<p><strong>Correction (2026-09-22)<\/strong>: A later check found that the content of this article conflicts with its sources.<br \/>\nThe article remains published to preserve the record.<\/p>\n<\/blockquote>\n<p><!-- lmw:facts --><\/p>\n<h2>At a Glance<\/h2>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Item<\/th>\n<th>Value<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Repository<\/td>\n<td><a href=\"https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF\">abenzerps\/Qwen-Image-2.1-Uncensored-GGUF<\/a><\/td>\n<\/tr>\n<tr>\n<td>Published<\/td>\n<td>2026-09-21<\/td>\n<\/tr>\n<tr>\n<td>License<\/td>\n<td>other<\/td>\n<\/tr>\n<tr>\n<td>Formats<\/td>\n<td>GGUF \/ safetensors<\/td>\n<\/tr>\n<tr>\n<td>Source type<\/td>\n<td>Unverified (not confirmed by a primary source)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Values determined by this site&#8217;s code at collection time. Dates are JST.<\/em><\/p>\n<p><!-- \/lmw:facts --><\/p>\n<p><em>Sample outputs are available on the <a href=\"https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF\">model card<\/a>.<\/em><\/p>\n<h2>Overview<\/h2>\n<p>It is reported that a quantized model repository <a href=\"https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF\">abenzerps\/Qwen-Image-2.1-Uncensored-GGUF<\/a> for &#8220;Qwen-Image-2.1&#8221;, a model supporting text-to-image generation and image editing, has been released. Note that this news is unverified information that has not been officially confirmed.<\/p>\n<p>This model is a text-to-image and image-editing model featuring capabilities such as generating regular images or transparent images (RGBA) from text prompts, editing existing images, and subject extraction. It is quantized into the GGUF format based on the base weights of the original model <a href=\"https:\/\/huggingface.co\/Qwen\/Qwen-Image-2.1\">Qwen\/Qwen-Image-2.1<\/a>, and is reported to have safety checkers and content filters removed. Execution in a local environment using ComfyUI and ComfyUI-GGUF is intended.<\/p>\n<h2>Specifications<\/h2>\n<p>Specifications based on documentation and public information of the original model are as follows:<\/p>\n<ul>\n<li>Parameter Count: 7B in the visual generation component (32 Single-Stream DiT layers)<\/li>\n<li>Architecture: Single-Stream DiT (mixed-granularity attention and prefix KV cache reuse structure)<\/li>\n<li>Output Specifications:<\/li>\n<li>Supports regular image generation and transparent image generation with an alpha channel (RGBA) &#8211; Supports transparent layer editing, subject extraction from photos, and identity-preserving editing of people or products using up to 10 reference images &#8211; Supports localized editing functions (circle designation, paint annotation, separate mask designation) &#8211; Supported aspect ratio and resolution examples: 1:1 (2048\u00d72048), 4:3 (2400\u00d71792), 3:4 (1792\u00d72400), 3:2 (2528\u00d71696), 2:3 (1696\u00d72528), 16:9 (2752\u00d71536), 9:16 (1536\u00d72752)<\/li>\n<li>Recommended Steps: 40 steps in the original model pipeline settings (num_inference_steps=40)<\/li>\n<li>Distribution Format: GGUF format (quantization types such as Q8_0, Q6_K, Q5_K_M, Q4_K_M, Q4_0), safetensors format (text encoder and VAE)<\/li>\n<li>License Terms: Qwen Research License<\/li>\n<\/ul>\n<h2>Performance and Quality<\/h2>\n<p>Information regarding the performance and quality of this model is reported based on the model card of the publisher and the information of the original model <a href=\"https:\/\/huggingface.co\/Qwen\/Qwen-Image-2.1\">Qwen\/Qwen-Image-2.1<\/a>. Although quantitative benchmark evaluation figures are not listed, the following characteristics regarding generation quality and behavior are indicated.<\/p>\n<p>In terms of the original model&#8217;s performance, it is reported to possess high generation quality despite a lightweight and efficient structure. In particular, it is claimed that improved accuracy in text rendering (typography) within images, appropriate lighting expression in portraits, and refined detail texture rendering have been achieved. Furthermore, as a quality strength, generation and editing are integrated into a single model, enabling the direct generation of images with an alpha channel from prompts that include background transparency instructions, as well as performing editing and generation without losing subject features even when combining up to 10 reference images.<\/p>\n<p>On the other hand, in the newly reported quantized version &#8220;abenzerps\/Qwen-Image-2.1-Uncensored-GGUF&#8221;, safety checkers and content filters are not incorporated. It is explained that it exhibits behavior that directly generates sensitive expressions, including adult content (NSFW), without performing prompt rejection or image blackout processing. The tendency and quality of the output deliverables are stated to depend entirely on the input prompt and execution environment.<\/p>\n<h2>Strengths and Use Cases<\/h2>\n<p>Based on the functionality of the original model Qwen\/Qwen-Image-2.1, this model is reported to be designed to support not only standard text-to-image generation but also multifaceted visual expression tasks. Specific areas of expertise and expected use cases are as follows:<\/p>\n<ul>\n<li><strong>Native Generation of Transparent Images (RGBA)<\/strong>: By specifying in the prompt that the background is transparent, you can directly generate images such as illustrations and sticker materials containing a transparent alpha channel.<\/li>\n<li><strong>Unified Image Editing and Subject Extraction<\/strong>: You can perform background changes, localized modifications, and extraction of specific subjects from photos on existing images using a single model.<\/li>\n<li><strong>Multi-Reference Generation Maintaining Identity<\/strong>: Supporting up to 10 reference images, it enables advanced editing such as generating a single group photo from multiple portrait photos without losing the features of a person&#8217;s face or product design.<\/li>\n<li><strong>Precise Rendering and Text Generation<\/strong>: It excels at accurate rendering of in-image text (typography) such as neon signs and signboards, appropriate lighting expression in portraits, and precise texture depiction.<\/li>\n<li><strong>Output in Various Aspect Ratios<\/strong>: In addition to square (1:1, 2048\u00d72048), it supports a wide variety of aspect ratios such as 4:3, 3:4, 3:2, 2:3, 16:9 (2752\u00d71536), and 9:16 (1536\u00d72752).<\/li>\n<\/ul>\n<p>Additionally, since content filters have been removed in this GGUF quantized version by abenzerps, it is stated to be provided for local engineers and creators who want to perform free creative work avoiding system prompt rejections and output blackouts. Furthermore, conversion to the GGUF format enables flexible execution in local environments with reduced memory usage.<\/p>\n<p><!-- lmw:hardware --><\/p>\n<h2>Hardware Requirements<\/h2>\n<p><strong>Estimated requirements (calculated by Local Model Watch)<\/strong> \u2014 7.1B parameters (taken from the base model Qwen\/Qwen-Image-2.1)<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Your VRAM<\/th>\n<th>Quantization<\/th>\n<th>File size<\/th>\n<th>Est. memory needed<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>8GB (RTX 4060 \/ 3060 Ti, etc.)<\/td>\n<td>Q6_K<\/td>\n<td>5.5GB<\/td>\n<td>6.6GB<\/td>\n<\/tr>\n<tr>\n<td>12GB (RTX 4070 \/ 3060 12GB, etc.)<\/td>\n<td>Q8_0<\/td>\n<td>7.1GB<\/td>\n<td>8.5GB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Memory estimates add a 20% runtime overhead (KV cache, etc.) to the actual size of the distributed files. Actual usage varies with context length, batch size and inference engine. These figures are computed by this site from file sizes, not published by the model&#8217;s authors. Compare with other models in our <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/vram-guide-en\/\">VRAM quick reference<\/a>. What the quantization names mean: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/glossary-quantization-en\/\">glossary<\/a>.<\/em><\/p>\n<p><!-- \/lmw:hardware --><\/p>\n<h2>How to Get It<\/h2>\n<p>Related files for this model are published in the Hugging Face repository <a href=\"https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF\">abenzerps\/Qwen-Image-2.1-Uncensored-GGUF<\/a>. License agreement procedures are reportedly not required for downloading (gated: false).<\/p>\n<p>To execute generation, an environment with <a href=\"https:\/\/github.com\/comfyanonymous\/ComfyUI\">ComfyUI<\/a> and <a href=\"https:\/\/github.com\/leejet\/ComfyUI-GGUF\">ComfyUI-GGUF<\/a> is required.<\/p>\n<h3>1. Downloading and Placing Necessary Files<\/h3>\n<p>Download the following files from the repository and place them in the designated ComfyUI directories:<\/p>\n<ul>\n<li><strong>Diffusion Model (GGUF)<\/strong>: Select one GGUF quantized file such as <code>qwen-image-2.1-Q4_K_M.gguf<\/code> and place it in <code>ComfyUI\/models\/diffusion_models\/<\/code>. The recommended quantization format is <code>Q4_K_M<\/code>, which offers excellent balance.<\/li>\n<li><strong>Text Encoder<\/strong>: Place <code>qwen3vl_8b_bf16.safetensors<\/code> or the memory-saving <code>qwen3vl_8b_int8_convrot.safetensors<\/code> in <code>ComfyUI\/models\/text_encoders\/<\/code>.<\/li>\n<li><strong>VAE<\/strong>: Place <code>qwen_image_2.1_vae_bf16.safetensors<\/code> in <code>ComfyUI\/models\/vae\/<\/code>.<\/li>\n<\/ul>\n<h3>2. ComfyUI Setup and Node Configuration<\/h3>\n<p>Utilize the custom node repository by <code>leejet<\/code>, which natively supports Qwen-Image 2.1.<\/p>\n<pre><code class=\"language-bash\">cd ComfyUI\/custom_nodes\ngit clone https:\/\/github.com\/leejet\/ComfyUI-GGUF\n<\/code><\/pre>\n<p><em>Note: If an &#8220;Unknown model architecture!&#8221; error occurs with older versions of <code>city96\/ComfyUI-GGUF<\/code>, it is reported that you need to update to the <code>leejet<\/code> version above or add <code>ModelQwenImage<\/code> to the conversion script.<\/em><\/p>\n<p>Configure the nodes on ComfyUI as follows:<\/p>\n<ul>\n<li><strong>Diffusion Model Node<\/strong>: Add the <strong><code>Unet Loader (GGUF)<\/code><\/strong> node and select the downloaded <code>.gguf<\/code> file.<\/li>\n<li><strong>Text Encoder Node<\/strong>: Add the standard <strong><code>CLIPLoader<\/code><\/strong> node and select <code>qwen3vl_8b_bf16.safetensors<\/code> (or the <code>int8<\/code> version). At this time, set <strong><code>type<\/code><\/strong> to <strong><code>qwen_image<\/code><\/strong>.<\/li>\n<li><strong>VAE Node<\/strong>: Add the standard <strong><code>VAELoader<\/code><\/strong> node and select <code>qwen_image_2.1_vae_bf16.safetensors<\/code>.<\/li>\n<\/ul>\n<p>When using the official workflow templates provided by Comfy-Org (for Text-to-Image or Image Edit), you can use them by replacing the standard <code>UNETLoader<\/code> node with the <code>Unet Loader (GGUF)<\/code> node.<\/p>\n<h3>3. Memory Optimization Settings<\/h3>\n<p>Recommended settings to increase memory efficiency without sacrificing generation speed during sampling are introduced in the model card:<\/p>\n<ul>\n<li><strong>Offloading Settings<\/strong>: It is recommended to keep the GGUF diffusion model in the GPU VRAM while placing and offloading the text encoder to system RAM (CPU). Since text encoding processing is executed only once per prompt, it is stated that VRAM usage can be significantly reduced with almost no impact on generation speed.<\/li>\n<li><strong>Recommended Configuration<\/strong>: A configuration combining <code>qwen-image-2.1-Q4_K_M.gguf<\/code> for the diffusion model and <code>qwen3vl_8b_int8_convrot.safetensors<\/code> for the text encoder is listed.<\/li>\n<li><strong>Low VRAM Mode<\/strong>: If errors due to insufficient VRAM (OOM) occur, it is recommended to specify <code>--lowvram<\/code> in the ComfyUI startup arguments.<\/li>\n<\/ul>\n<p><!-- lmw:related --><\/p>\n<h2>Related Articles<\/h2>\n<ul>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/21\/qwen-image-2-1-gguf-released\/\">Qwen-Image-2.1-GGUF Released: Local Image Generation<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:related --><\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF\">https:\/\/huggingface.co\/abenzerps\/Qwen-Image-2.1-Uncensored-GGUF<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/Qwen\/Qwen-Image-2.1\">https:\/\/huggingface.co\/Qwen\/Qwen-Image-2.1<\/a><\/li>\n<\/ul>\n<p><!-- lmw:updates --><\/p>\n<h2>Update History<\/h2>\n<ul>\n<li>2026-09-22: A discrepancy with the sources was found; a correction notice was added at the top of the article.<\/li>\n<\/ul>\n<p><!-- \/lmw:updates --><\/p>\n<blockquote>\n<p><strong>This article contains unverified information.<\/strong> We will append an update note once it is confirmed by a primary source.<\/p>\n<\/blockquote>\n","protected":false},"excerpt":{"rendered":"<p>An unofficial uncensored GGUF quantization of Qwen-Image-2.1 has been released for local image generation and editing using ComfyUI.<\/p>\n","protected":false},"author":1,"featured_media":2342,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[203],"tags":[1852,677,163,1835,1565,377,753],"class_list":["post-2343","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-image-video-and-audio","tag-abenzerps-en","tag-comfyui-en","tag-gguf-en","tag-qwen-image-2-1-en","tag-unverified","tag--en"],"lang":"en","translations":{"en":2343,"ja":2341},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/2343","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/comments?post=2343"}],"version-history":[{"count":1,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/2343\/revisions"}],"predecessor-version":[{"id":2450,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/2343\/revisions\/2450"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media\/2342"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=2343"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/categories?post=2343"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/tags?post=2343"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}